How to Prepare for a Senior Data Engineer Interview in 30 Days
A 30-day study plan: Spark internals (week 1), Kafka + streaming (week 2), system design (week 3), SQL + case study (week 4).
Get one deep-dive article every Saturday. No spam. Unsubscribe anytime.
Senior data engineer interviews at companies like Flipkart, Swiggy, Zepto, and CRED are technically demanding. They're not testing whether you can write a groupBy — they're testing whether you understand why things work the way they do.
What They Actually Test
Spark internals (not just API). How does AQE work? What happens during a shuffle? How do you fix data skew?
Kafka + streaming design. Consumer group semantics, exactly-once, consumer lag.
SQL and data modeling. Window functions, SCD Type 2, recursive CTEs.
System design. Lambda vs Kappa, data mesh, designing a data platform for 100TB/day.
Week 1: Spark Internals
Day 1–2: Read Spark's execution model — stages, tasks, shuffle.
Day 3–4: Enable AQE locally, run a query with/without, read EXPLAIN output.
Day 5–7: Practice explaining data skew. What are the five ways to fix it?
// Week 1 practice: explain what happens
val df = spark.read.parquet("s3://events/")
.filter($"event_type" === "purchase")
.join(users, "user_id")
.groupBy($"country")
.agg(sum($"amount").as("revenue"))
// How many stages? Where is shuffle? If users is 50MB, what optimization?Get unlimited access to all Career tutorials & animated diagrams
1,000+ animated tutorials · 100+ AI tools · Mock interviews · Certifications
Week 2: Kafka + Streaming
Day 8–10: Consumer groups — draw partition assignment for 6 partitions, 3 consumers.
Day 11–12: Exactly-once: idempotent producers, transactions, and where EOS breaks.
Day 13–14: System design: "Design a real-time pipeline at 100K events/sec."
Week 3: System Design
Day 15–17: Lambda vs Kappa — build an argument for each.
Day 18–19: Lakehouse: Delta vs Iceberg.
Day 20–21: Data mesh — what does a self-serve platform look like?
Use STAR format: Situation, Task, Action, Result.
Week 4: SQL + Case Study
Day 22–23: Window functions from memory: RANK, ROW_NUMBER, LAG.
Day 24–25: SCD Type 2 in dbt with snapshots AND in pure SQL with MERGE.
Day 26–28: Case study: "Revenue report shows ₹50L less than finance — debug it."
Day 29–30: Mock interviews. Record yourself. The way you explain matters as much as the answer.
Get unlimited access to all Career tutorials & animated diagrams
1,000+ animated tutorials · 100+ AI tools · Mock interviews · Certifications
Get unlimited access to all Career tutorials & animated diagrams
1,000+ animated tutorials · 100+ AI tools · Mock interviews · Certifications